Statistical analysis of living habits of individual persons and households
Jasmine Lundgren
Ha Do
Roosa Muranen
Authors
Load data
We'll start the data analysis process by importing necessary libraries and loading the data. We will only import the columns which are examined
in the analysis and we also will replace question mark values by NaN values.
household ID member ID Day of week Sex Living environment Age group Cooking Washing dishes Listening to radio Phonecall Museum Library
0 50007 2 2 2 3.0 6 0 20 0 0 2.0 1.0
1 50009 1 1 2 1.0 7 40 0 0 0 2.0 1.0
2 50015 1 1 1 3.0 8 10 0 10 0 2.0 1.0
3 50032 2 1 1 2.0 8 0 10 0 0 2.0 2.0
4 50033 1 1 2 1.0 8 02:10 00:20 00:00 00:00 2.0 2.0
# Import libaries are needed for data processing
import pandas as pd
import numpy as np
import scipy.stats as s
from sklearn.preprocessing import StandardScaler
from kmodes.kprototypes import KPrototypes
import matplotlib.pyplot as plt
import seaborn as sns
import re
import statsmodels.stats as ss
import statsmodels.stats.multitest as ssm
#Load and read dataset from a file called habits.data
file_path = 'habits.data'
data = pd.read_csv(file_path, sep=';', skiprows=1, usecols=[0, 1, 2, 3, 4, 5, 7, 8, 16, 18, 21, 22], names=['household ID', 'member ID', 'Day of week', 'Sex', 'Living environment', 'Age group',
'Cooking', 'Washing dishes', 'Listening to radio', 'Phonecall',
'Museum', 'Library'], na_values=['?'])
data.head()
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 1/36
Data preparation and exploration
We classify data into categorical and numeric variables since they have different data types. Columns such as sex, living environment, age
group, museum, and library are classi ed as categorical because they do not have continuous values like numerical columns. These variables
represent different demographic information, such as gender, living environment, age group, and has a person visited a museum or a library.
Meanwhile, numerical data includes how an individual has spent time on different activities such as cooking, washing dishes, listening to radio,
and making phone calls.
# Define numerical_columns
numerical_columns = ['Cooking', 'Washing dishes', 'Listening to radio', 'Phonecall']
# Define categorial columns
categorical_columns = ['Day of week', 'Sex', 'Living environment', 'Age group', 'Museum', 'Library']
As we start to inspect the data, we can see, that all of the household id values are unique. That leads us to a conclusion that there's only one
person per household in our data.
# There seems to be no duplicate household_ids so we assume there's only one person per household
print(len(data))
print(len(pd.unique(data['household ID'])))
393
393
When we check for missing values, we can see, that there's 44 rows containing NaN values. After calculating missing values for all variables, it
seems that most of the missing values belong to the museum and library activities.
# Check if any missing values
data.isna().any(axis=1).sum()
44
# Check missing values by variable
data.isnull().sum()
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 2/36
We classify data into categorical and numeric variables since they have different data types. Columns such as sex, living environment, age
group, museum, and library are classi ed as categorical because they do not have continuous values like numerical columns. These variables
represent different demographic information, such as gender, living environment, age group, and has a person visited a museum or a library.
Meanwhile, numerical data includes how an individual has spent time on different activities such as cooking, washing dishes, listening to radio,
and making phone calls.
# Define numerical_columns
numerical_columns = ['Cooking', 'Washing dishes', 'Listening to radio', 'Phonecall']
# Define categorial columns
categorical_columns = ['Day of week', 'Sex', 'Living environment', 'Age group', 'Museum', 'Library']
As we start to inspect the data, we can see, that all of the household id values are unique. That leads us to a conclusion that there's only one
person per household in our data.
# There seems to be no duplicate household_ids so we assume there's only one person per household
print(len(data))
print(len(pd.unique(data['household ID'])))
393
393
When we check for missing values, we can see, that there's 44 rows containing NaN values. After calculating missing values for all variables, it
seems that most of the missing values belong to the museum and library activities.
# Check if any missing values
data.isna().any(axis=1).sum()
44
# Check missing values by variable
data.isnull().sum()
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 2/36
0
household ID 0
member ID 0
Day of week 0
Sex 0
Living environment 1
Age group 0
Cooking 5
Washing dishes 4
Listening to radio 4
Phonecall 7
Museum 24
Library 23
dtype: int64
We decided to drop the rows with null values since the number of rows with them isn't too high and most missing values are in the museum and
library columns, which can only get yes or no values, so replacing a null value with the mean could sway the results.
household ID member ID Day of week Sex Living environment Age group Cooking Washing dishes Listening to radio Phonecall Museum Library
0 50007 2 2 2 3.0 6 0 20 0 0 2.0 1.0
1 50009 1 1 2 1.0 7 40 0 0 0 2.0 1.0
2 50015 1 1 1 3.0 8 10 0 10 0 2.0 1.0
3 50032 2 1 1 2.0 8 0 10 0 0 2.0 2.0
4 50033 1 1 2 1.0 8 02:10 00:20 00:00 00:00 2.0 2.0
# Drop rows have NaN values
data.dropna(inplace=True)
data.head()
After dealing with the missing values, we observe the categorical variables for erronous data. By observing the printed values we can see that
there's some erronous data with museum and library variables.
# Check categorical variables for odd values before converting
print((data['Day of week'] < 1).value_counts())
print((data['Day of week'] > 2).value_counts())
print((data['Sex'] < 1).value_counts())
print((data['Sex'] > 2).value_counts())
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 3/36
household ID 0
member ID 0
Day of week 0
Sex 0
Living environment 1
Age group 0
Cooking 5
Washing dishes 4
Listening to radio 4
Phonecall 7
Museum 24
Library 23
dtype: int64
We decided to drop the rows with null values since the number of rows with them isn't too high and most missing values are in the museum and
library columns, which can only get yes or no values, so replacing a null value with the mean could sway the results.
household ID member ID Day of week Sex Living environment Age group Cooking Washing dishes Listening to radio Phonecall Museum Library
0 50007 2 2 2 3.0 6 0 20 0 0 2.0 1.0
1 50009 1 1 2 1.0 7 40 0 0 0 2.0 1.0
2 50015 1 1 1 3.0 8 10 0 10 0 2.0 1.0
3 50032 2 1 1 2.0 8 0 10 0 0 2.0 2.0
4 50033 1 1 2 1.0 8 02:10 00:20 00:00 00:00 2.0 2.0
# Drop rows have NaN values
data.dropna(inplace=True)
data.head()
After dealing with the missing values, we observe the categorical variables for erronous data. By observing the printed values we can see that
there's some erronous data with museum and library variables.
# Check categorical variables for odd values before converting
print((data['Day of week'] < 1).value_counts())
print((data['Day of week'] > 2).value_counts())
print((data['Sex'] < 1).value_counts())
print((data['Sex'] > 2).value_counts())
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 3/36
print((data['Living environment'] < 1).value_counts())
print((data['Living environment'] > 3).value_counts())
print((data['Age group'] < 1).value_counts())
print((data['Age group'] > 9).value_counts())
print((data['Museum'] < 1).value_counts())
print((data['Museum'] > 2).value_counts())
print((data['Library']< 1).value_counts())
print((data['Library'] > 2).value_counts())
Day of week
False 349
Name: count, dtype: int64
Day of week
False 349
Name: count, dtype: int64
Sex
False 349
Name: count, dtype: int64
Sex
False 349
Name: count, dtype: int64
Living environment
False 349
Name: count, dtype: int64
Living environment
False 349
Name: count, dtype: int64
Age group
False 349
Name: count, dtype: int64
Age group
False 349
Name: count, dtype: int64
Museum
False 345
True 4
Name: count, dtype: int64
Museum
False 347
True 2
Name: count, dtype: int64
Library
False 343
True 6
Name: count, dtype: int64
Library
False 349
Name: count, dtype: int64
We decided to drop the erronous rows for the same reasons as with the whole dataset.
# Some erronous data on visited_library and visited_museum columns
# Drop the rows
index = data[(data['Museum'] < 1)].index
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 4/36
print((data['Living environment'] > 3).value_counts())
print((data['Age group'] < 1).value_counts())
print((data['Age group'] > 9).value_counts())
print((data['Museum'] < 1).value_counts())
print((data['Museum'] > 2).value_counts())
print((data['Library']< 1).value_counts())
print((data['Library'] > 2).value_counts())
Day of week
False 349
Name: count, dtype: int64
Day of week
False 349
Name: count, dtype: int64
Sex
False 349
Name: count, dtype: int64
Sex
False 349
Name: count, dtype: int64
Living environment
False 349
Name: count, dtype: int64
Living environment
False 349
Name: count, dtype: int64
Age group
False 349
Name: count, dtype: int64
Age group
False 349
Name: count, dtype: int64
Museum
False 345
True 4
Name: count, dtype: int64
Museum
False 347
True 2
Name: count, dtype: int64
Library
False 343
True 6
Name: count, dtype: int64
Library
False 349
Name: count, dtype: int64
We decided to drop the erronous rows for the same reasons as with the whole dataset.
# Some erronous data on visited_library and visited_museum columns
# Drop the rows
index = data[(data['Museum'] < 1)].index
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 4/36
data.drop(index, inplace=True)
index = data[(data['Museum'] > 2)].index
data.drop(index, inplace=True)
index = data[(data['Library'] < 1)].index
data.drop(index, inplace=True)
print((data['Museum'] < 1).value_counts())
print((data['Museum'] > 2).value_counts())
print((data['Library'] < 1).value_counts())
Museum
False 343
Name: count, dtype: int64
Museum
False 343
Name: count, dtype: int64
Library
False 343
Name: count, dtype: int64
When we have all set with the categorical variables in the data, we'll rename the values to be more descriptive and change the variables'
datatype to category.
household ID member ID Day of week Sex Living environment Age group Cooking Washing dishes Listening to radio Phonecall Museum Library
0 50007 2 weekend Female Rural 45-54 years old 0 20 0 0 No Yes
1 50009 1 working day Female City 55-64 years old 40 0 0 0 No Yes
2 50015 1 working day Male Rural 65-74 years old 10 0 10 0 No Yes
3 50032 2 working day Male Municipality 65-74 years old 0 10 0 0 No No
4 50033 1 working day Female City 65-74 years old 02:10 00:20 00:00 00:00 No No
# Map encoded values to their corresponding labels
data['Day of week'] = data['Day of week'].map({1: 'working day', 2: 'weekend'})
data['Sex'] = data['Sex'].map({1: 'Male', 2: 'Female'})
data['Living environment'] = data['Living environment'].map({1: 'City', 2: 'Municipality', 3: 'Rural'})
data['Age group'] = data['Age group'].map({
1: '10-14 years old', 2: '15-19 years old', 3: '20-24 years old',
4: '25-34 years old', 5: '35-44 years old', 6: '45-54 years old',
7: '55-64 years old', 8: '65-74 years old', 9: '75+ years old'
})
data['Museum'] = data['Museum'].map({1: 'Yes', 2: 'No'})
data['Library'] = data['Library'].map({1: 'Yes', 2: 'No'})
# Convert all categorial columns to category datatype
for col in categorical_columns:
data[col] = data[col].astype('category')
data.head()
While glancing the rst few rows of the data, we notice that there are some values in the activity variables that are in "hh:mm" format instead of
the expected minute format. We have to convert those values to minute format, so we created a function for that and applied it to the activity
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 5/36
index = data[(data['Museum'] > 2)].index
data.drop(index, inplace=True)
index = data[(data['Library'] < 1)].index
data.drop(index, inplace=True)
print((data['Museum'] < 1).value_counts())
print((data['Museum'] > 2).value_counts())
print((data['Library'] < 1).value_counts())
Museum
False 343
Name: count, dtype: int64
Museum
False 343
Name: count, dtype: int64
Library
False 343
Name: count, dtype: int64
When we have all set with the categorical variables in the data, we'll rename the values to be more descriptive and change the variables'
datatype to category.
household ID member ID Day of week Sex Living environment Age group Cooking Washing dishes Listening to radio Phonecall Museum Library
0 50007 2 weekend Female Rural 45-54 years old 0 20 0 0 No Yes
1 50009 1 working day Female City 55-64 years old 40 0 0 0 No Yes
2 50015 1 working day Male Rural 65-74 years old 10 0 10 0 No Yes
3 50032 2 working day Male Municipality 65-74 years old 0 10 0 0 No No
4 50033 1 working day Female City 65-74 years old 02:10 00:20 00:00 00:00 No No
# Map encoded values to their corresponding labels
data['Day of week'] = data['Day of week'].map({1: 'working day', 2: 'weekend'})
data['Sex'] = data['Sex'].map({1: 'Male', 2: 'Female'})
data['Living environment'] = data['Living environment'].map({1: 'City', 2: 'Municipality', 3: 'Rural'})
data['Age group'] = data['Age group'].map({
1: '10-14 years old', 2: '15-19 years old', 3: '20-24 years old',
4: '25-34 years old', 5: '35-44 years old', 6: '45-54 years old',
7: '55-64 years old', 8: '65-74 years old', 9: '75+ years old'
})
data['Museum'] = data['Museum'].map({1: 'Yes', 2: 'No'})
data['Library'] = data['Library'].map({1: 'Yes', 2: 'No'})
# Convert all categorial columns to category datatype
for col in categorical_columns:
data[col] = data[col].astype('category')
data.head()
While glancing the rst few rows of the data, we notice that there are some values in the activity variables that are in "hh:mm" format instead of
the expected minute format. We have to convert those values to minute format, so we created a function for that and applied it to the activity
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 5/36
variables. In addition, we changed the variables' datatype to int64.
# Convert time like 00:20 in numerical_columns into numerical format
# Regex for hh:mm
pattern = re.compile('^([0-1]?[0-9]|2[0-3]):[0-5][0-9]$')
# Create a function to convert
def convert(value):
if isinstance(value, str) and pattern.match(value):
hours, minutes = map(int, value.split(':'))
return hours * 60 + minutes
return int(value)
# Apply convestion into numerical_columns
for col in numerical_columns:
data[col] = data[col].apply(convert).astype('int64')
data.info()
<class 'pandas.core.frame.DataFrame'>
Index: 343 entries, 0 to 391
Data columns (total 12 columns):
# Column Non-Null Count Dtype
--- ------ -------------- -----
0 household ID 343 non-null int64
1 member ID 343 non-null int64
2 Day of week 343 non-null category
3 Sex 343 non-null category
4 Living environment 343 non-null category
5 Age group 343 non-null category
6 Cooking 343 non-null int64
7 Washing dishes 343 non-null int64
8 Listening to radio 343 non-null int64
9 Phonecall 343 non-null int64
10 Museum 343 non-null category
11 Library 343 non-null category
dtypes: category(6), int64(6)
memory usage: 21.7 KB
After the numerical variables had been converted to the suitable format and datatype, we inspect the variables for erronous values. From the
output we could see that there weren't any radical values.
# Check numerical variables for odd values
print((data['Cooking'] < 0).value_counts())
print((data['Washing dishes'] < 0).value_counts())
print((data['Listening to radio'] < 0).value_counts())
print((data['Phonecall'] < 0).value_counts())
print((data['Cooking'] > 1000).value_counts())
print((data['Washing dishes'] > 1000).value_counts())
print((data['Listening to radio'] > 1000).value_counts())
print((data['Phonecall'] > 1000).value_counts())
Cooking
False 343
Name: count, dtype: int64
Washing dishes
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 6/36
# Convert time like 00:20 in numerical_columns into numerical format
# Regex for hh:mm
pattern = re.compile('^([0-1]?[0-9]|2[0-3]):[0-5][0-9]$')
# Create a function to convert
def convert(value):
if isinstance(value, str) and pattern.match(value):
hours, minutes = map(int, value.split(':'))
return hours * 60 + minutes
return int(value)
# Apply convestion into numerical_columns
for col in numerical_columns:
data[col] = data[col].apply(convert).astype('int64')
data.info()
<class 'pandas.core.frame.DataFrame'>
Index: 343 entries, 0 to 391
Data columns (total 12 columns):
# Column Non-Null Count Dtype
--- ------ -------------- -----
0 household ID 343 non-null int64
1 member ID 343 non-null int64
2 Day of week 343 non-null category
3 Sex 343 non-null category
4 Living environment 343 non-null category
5 Age group 343 non-null category
6 Cooking 343 non-null int64
7 Washing dishes 343 non-null int64
8 Listening to radio 343 non-null int64
9 Phonecall 343 non-null int64
10 Museum 343 non-null category
11 Library 343 non-null category
dtypes: category(6), int64(6)
memory usage: 21.7 KB
After the numerical variables had been converted to the suitable format and datatype, we inspect the variables for erronous values. From the
output we could see that there weren't any radical values.
# Check numerical variables for odd values
print((data['Cooking'] < 0).value_counts())
print((data['Washing dishes'] < 0).value_counts())
print((data['Listening to radio'] < 0).value_counts())
print((data['Phonecall'] < 0).value_counts())
print((data['Cooking'] > 1000).value_counts())
print((data['Washing dishes'] > 1000).value_counts())
print((data['Listening to radio'] > 1000).value_counts())
print((data['Phonecall'] > 1000).value_counts())
Cooking
False 343
Name: count, dtype: int64
Washing dishes
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 6/36
False 343
Name: count, dtype: int64
Listening to radio
False 343
Name: count, dtype: int64
Phonecall
False 343
Name: count, dtype: int64
Cooking
False 343
Name: count, dtype: int64
Washing dishes
False 343
Name: count, dtype: int64
Listening to radio
False 343
Name: count, dtype: int64
Phonecall
False 343
Name: count, dtype: int64
We also want to observe the maximum and minimum values for the activity variables.
The highest value for time spent on an activity is listening to radio. While the number seems to be quite high, by converting it to hours (~8,3 h)
we can see that it's a plausible value since e.g. a person could listen to radio while working or use a radio as a background noise. The second
highest value is for cooking which also seems extraordinary (6 h) but can be considered as a possibly correct value.
The lowest value for all activities is 0. It's a plausible value for each activity while it may seem unlikely.
0
Cooking 360
Washing dishes 110
Listening to radio 500
Phonecall 110
dtype: int64
# What are maximum values for each activity
data[numerical_columns].max()
0
Cooking 0
Washing dishes 0
Listening to radio 0
Phonecall 0
dtype: int64
# What are minimum values for each activity
data[numerical_columns].min()
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 7/36
Name: count, dtype: int64
Listening to radio
False 343
Name: count, dtype: int64
Phonecall
False 343
Name: count, dtype: int64
Cooking
False 343
Name: count, dtype: int64
Washing dishes
False 343
Name: count, dtype: int64
Listening to radio
False 343
Name: count, dtype: int64
Phonecall
False 343
Name: count, dtype: int64
We also want to observe the maximum and minimum values for the activity variables.
The highest value for time spent on an activity is listening to radio. While the number seems to be quite high, by converting it to hours (~8,3 h)
we can see that it's a plausible value since e.g. a person could listen to radio while working or use a radio as a background noise. The second
highest value is for cooking which also seems extraordinary (6 h) but can be considered as a possibly correct value.
The lowest value for all activities is 0. It's a plausible value for each activity while it may seem unlikely.
0
Cooking 360
Washing dishes 110
Listening to radio 500
Phonecall 110
dtype: int64
# What are maximum values for each activity
data[numerical_columns].max()
0
Cooking 0
Washing dishes 0
Listening to radio 0
Phonecall 0
dtype: int64
# What are minimum values for each activity
data[numerical_columns].min()
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 7/36
Next for getting a basic overview of the data, we calculate some sums based on the individuals' demographic features.
While there weren't any drastic difference in the amount of individuals for each sex, we could see that there is slightly more females than males
in the dataset.
# How many representatives for each sex
print('Male:')
print(data[data.Sex == 'Male'].shape[0])
print('Female:')
print(data[data.Sex == 'Female'].shape[0])
Male:
153
Female:
190
<Axes: xlabel='Sex'>
data['Sex'].value_counts().plot(kind='bar')
When it comes to the individuals' living environment, we could witness that most of the individuals live in the city and the least in the
municipality.
# How many representatives for each living environment
print('City:')
print(data[data['Living environment'] == 'City'].shape[0])
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 8/36
While there weren't any drastic difference in the amount of individuals for each sex, we could see that there is slightly more females than males
in the dataset.
# How many representatives for each sex
print('Male:')
print(data[data.Sex == 'Male'].shape[0])
print('Female:')
print(data[data.Sex == 'Female'].shape[0])
Male:
153
Female:
190
<Axes: xlabel='Sex'>
data['Sex'].value_counts().plot(kind='bar')
When it comes to the individuals' living environment, we could witness that most of the individuals live in the city and the least in the
municipality.
# How many representatives for each living environment
print('City:')
print(data[data['Living environment'] == 'City'].shape[0])
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 8/36
print('Municipality:')
print(data[data['Living environment'] == 'Municipality'].shape[0])
print('Rural area:')
print(data[data['Living environment'] == 'Rural'].shape[0])
City:
212
Municipality:
55
Rural area:
76
<Axes: xlabel='Living environment'>
data['Living environment'].value_counts().plot(kind='bar')
When observing the number of individuals in each age group, we can state that most of the individuals in the data are over 45 years old. There
seems to be no representatives of the age groups between 10-19 year olds. The largest age groups in the data are between 45-64 year olds.
# How many representatives for each age group
print('10-14 years old:')
print(data[data['Age group'] == '10-14 years old'].shape[0])
print('15-19 years old:')
print(data[data['Age group'] == '15-19 years old'].shape[0])
print('20-24 years old:')
print(data[data['Age group'] == '20-24 years old'].shape[0])
print('25-34 years old:')
print(data[data['Age group'] == '25-34 years old'].shape[0])
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 9/36
print(data[data['Living environment'] == 'Municipality'].shape[0])
print('Rural area:')
print(data[data['Living environment'] == 'Rural'].shape[0])
City:
212
Municipality:
55
Rural area:
76
<Axes: xlabel='Living environment'>
data['Living environment'].value_counts().plot(kind='bar')
When observing the number of individuals in each age group, we can state that most of the individuals in the data are over 45 years old. There
seems to be no representatives of the age groups between 10-19 year olds. The largest age groups in the data are between 45-64 year olds.
# How many representatives for each age group
print('10-14 years old:')
print(data[data['Age group'] == '10-14 years old'].shape[0])
print('15-19 years old:')
print(data[data['Age group'] == '15-19 years old'].shape[0])
print('20-24 years old:')
print(data[data['Age group'] == '20-24 years old'].shape[0])
print('25-34 years old:')
print(data[data['Age group'] == '25-34 years old'].shape[0])
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 9/36
print('35-44 years old:')
print(data[data['Age group'] == '35-44 years old'].shape[0])
print('45-54 years old:')
print(data[data['Age group'] == '45-54 years old'].shape[0])
print('55-64 years old:')
print(data[data['Age group'] == '55-64 years old'].shape[0])
print('65-74 years old:')
print(data[data['Age group'] == '65-74 years old'].shape[0])
print('75+ years old:')
print(data[data['Age group'] == '75+ years old'].shape[0])
10-14 years old:
0
15-19 years old:
0
20-24 years old:
11
25-34 years old:
48
35-44 years old:
52
45-54 years old:
80
55-64 years old:
80
65-74 years old:
47
75+ years old:
25
data['Age group'].value_counts(sort=False).plot(kind='bar')
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 10/36
print(data[data['Age group'] == '35-44 years old'].shape[0])
print('45-54 years old:')
print(data[data['Age group'] == '45-54 years old'].shape[0])
print('55-64 years old:')
print(data[data['Age group'] == '55-64 years old'].shape[0])
print('65-74 years old:')
print(data[data['Age group'] == '65-74 years old'].shape[0])
print('75+ years old:')
print(data[data['Age group'] == '75+ years old'].shape[0])
10-14 years old:
0
15-19 years old:
0
20-24 years old:
11
25-34 years old:
48
35-44 years old:
52
45-54 years old:
80
55-64 years old:
80
65-74 years old:
47
75+ years old:
25
data['Age group'].value_counts(sort=False).plot(kind='bar')
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 10/36
<Axes: xlabel='Age group'>
1. Characterising individuals present in the data
One of the descriptive statistic methods we use is K-Prototypes clustering, aiming to group similar individuals into three clusters based on the
individual's sex, age group, living environment, how they spent time on the different activities and have they visited library or museum in the past
year.
K-prototypes combine two clustering algorithms: K-means that groups data points together based on their similarity in distance to the centroid
of their respective clusters and K-modes that calculates the number of mismatches between categorical values [4], [7].
K-prototypes was thought to be the most suitable option to apply to our dataset, since it contains both categorical and numerical variables.
# Define categorical columns for clustering
categorical_columns_clus = ['Sex', 'Living environment', 'Age group', 'Museum', 'Library']
# Combine all columns for clustering
all_columns = categorical_columns_clus + numerical_columns
# Define the indices of categorical columns in the DataFrame
categorical_indices = [all_columns.index(col) for col in categorical_columns_clus]
# Initialize K prototype model
kproto = KPrototypes(n_clusters=3, random_state=42)
clusters = kproto.fit_predict(data[all_columns], categorical=categorical_indices)
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 11/36
1. Characterising individuals present in the data
One of the descriptive statistic methods we use is K-Prototypes clustering, aiming to group similar individuals into three clusters based on the
individual's sex, age group, living environment, how they spent time on the different activities and have they visited library or museum in the past
year.
K-prototypes combine two clustering algorithms: K-means that groups data points together based on their similarity in distance to the centroid
of their respective clusters and K-modes that calculates the number of mismatches between categorical values [4], [7].
K-prototypes was thought to be the most suitable option to apply to our dataset, since it contains both categorical and numerical variables.
# Define categorical columns for clustering
categorical_columns_clus = ['Sex', 'Living environment', 'Age group', 'Museum', 'Library']
# Combine all columns for clustering
all_columns = categorical_columns_clus + numerical_columns
# Define the indices of categorical columns in the DataFrame
categorical_indices = [all_columns.index(col) for col in categorical_columns_clus]
# Initialize K prototype model
kproto = KPrototypes(n_clusters=3, random_state=42)
clusters = kproto.fit_predict(data[all_columns], categorical=categorical_indices)
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 11/36
household ID member ID Day of week Sex Living environment Age group Cooking Washing dishes Listening to radio Phonecall Museum Library Cluster
0 50007 2 weekend Female Rural 45-54 years old 0 20 0 0 No Yes 1
1 50009 1 working day Female City 55-64 years old 40 0 0 0 No Yes 1
2 50015 1 working day Male Rural 65-74 years old 10 0 10 0 No Yes 1
3 50032 2 working day Male Municipality 65-74 years old 0 10 0 0 No No 1
4 50033 1 working day Female City 65-74 years old 130 20 0 0 No No 2
# Add cluster to data
data['Cluster'] = clusters
data.head()
# Summarize demographic and activities in each cluster
cluster_summary = data.groupby('Cluster').agg({
'Age group': lambda x: x.mode()[0], # Most common age group in each cluster
'Sex': lambda x: x.mode()[0], # Most common sex in each cluster
'Living environment': lambda x: x.mode()[0], # Most common living environment
'Cooking': 'mean',
'Washing dishes': 'mean',
'Listening to radio': 'mean',
'Phonecall': 'mean',
'Museum': lambda x: x.mode()[0], # Most common visit or not visit in each cluster
'Library': lambda x: x.mode()[0] # Most common visit or not visit in each cluster
})
print("Cluster Summary:\n", cluster_summary)
Cluster Summary:
Age group Sex Living environment Cooking \
Cluster
0 55-64 years old Male City 41.666667
1 55-64 years old Male City 15.506608
2 65-74 years old Female City 80.909091
Washing dishes Listening to radio Phonecall Museum Library
Cluster
0 10.000000 235.000000 0.000000 No No
1 5.286344 4.757709 4.185022 No Yes
2 28.000000 4.090909 11.363636 No Yes
# Visualize the distribution of Age groups within each cluster
plt.figure(figsize=(12, 6))
sns.countplot(data=data, x='Age group', hue='Cluster')
plt.title('Age Group Distribution Throughout Cluster')
plt.show()
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 12/36
0 50007 2 weekend Female Rural 45-54 years old 0 20 0 0 No Yes 1
1 50009 1 working day Female City 55-64 years old 40 0 0 0 No Yes 1
2 50015 1 working day Male Rural 65-74 years old 10 0 10 0 No Yes 1
3 50032 2 working day Male Municipality 65-74 years old 0 10 0 0 No No 1
4 50033 1 working day Female City 65-74 years old 130 20 0 0 No No 2
# Add cluster to data
data['Cluster'] = clusters
data.head()
# Summarize demographic and activities in each cluster
cluster_summary = data.groupby('Cluster').agg({
'Age group': lambda x: x.mode()[0], # Most common age group in each cluster
'Sex': lambda x: x.mode()[0], # Most common sex in each cluster
'Living environment': lambda x: x.mode()[0], # Most common living environment
'Cooking': 'mean',
'Washing dishes': 'mean',
'Listening to radio': 'mean',
'Phonecall': 'mean',
'Museum': lambda x: x.mode()[0], # Most common visit or not visit in each cluster
'Library': lambda x: x.mode()[0] # Most common visit or not visit in each cluster
})
print("Cluster Summary:\n", cluster_summary)
Cluster Summary:
Age group Sex Living environment Cooking \
Cluster
0 55-64 years old Male City 41.666667
1 55-64 years old Male City 15.506608
2 65-74 years old Female City 80.909091
Washing dishes Listening to radio Phonecall Museum Library
Cluster
0 10.000000 235.000000 0.000000 No No
1 5.286344 4.757709 4.185022 No Yes
2 28.000000 4.090909 11.363636 No Yes
# Visualize the distribution of Age groups within each cluster
plt.figure(figsize=(12, 6))
sns.countplot(data=data, x='Age group', hue='Cluster')
plt.title('Age Group Distribution Throughout Cluster')
plt.show()
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 12/36
# Calulate average time spent on activities for each cluster
activity_means = data.groupby('Cluster')[numerical_columns].mean().reset_index()
# Plot average time spent activity throughout each cluster
activity_means.set_index('Cluster').plot(kind='bar', figsize=(10, 6))
plt.title('Average Time Spent on Activities by Cluster')
plt.ylabel('Average Time (minutes)')
plt.xlabel('Cluster')
plt.legend(title='Activity', loc='upper right')
plt.tight_layout()
plt.show()
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 13/36
activity_means = data.groupby('Cluster')[numerical_columns].mean().reset_index()
# Plot average time spent activity throughout each cluster
activity_means.set_index('Cluster').plot(kind='bar', figsize=(10, 6))
plt.title('Average Time Spent on Activities by Cluster')
plt.ylabel('Average Time (minutes)')
plt.xlabel('Cluster')
plt.legend(title='Activity', loc='upper right')
plt.tight_layout()
plt.show()
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 13/36
Based on the cluster summary, the clustering process describes three different groups (cluster 0, cluster 1, cluster 2). Each cluster represents
individuals with speci c characterics based on their demographics and time spent on different activities. The cluster summary interprets age
group, sex, living environment, and time spent on various activities, such as cooking, washing dishes, listening to radio and phonecalls.
This cluster primarily includes people in the 55-64 years old age group, with mostly males living in the city
Activities: On average, cooking and listening to radio take up the most time each day (41.6 and 235 minutes, respectively). The third most
time is spent on washing dishes which was 10 minutes per day on average. Lastly, museum and libary are two activities which were not
attractive to this group and they did not vist in the past 12 months.
Cluster 1
This cluster primarily includes people in the 45-54 years old age group, with mostly males living in the city
Activities: On average, cooking and washing dishes take up the most time each day (16.05 and 5.3 minutes, respectively). The next most
time is spent on listenting to radio and phonecalls which accounts for about 4.6 and 4 minutes per day on average. Visiting libraries is a
common interest among this group whereas visiting museums has not been in their interests in the past 12 months.
Cluster 2
Cluster 0
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 14/36
individuals with speci c characterics based on their demographics and time spent on different activities. The cluster summary interprets age
group, sex, living environment, and time spent on various activities, such as cooking, washing dishes, listening to radio and phonecalls.
This cluster primarily includes people in the 55-64 years old age group, with mostly males living in the city
Activities: On average, cooking and listening to radio take up the most time each day (41.6 and 235 minutes, respectively). The third most
time is spent on washing dishes which was 10 minutes per day on average. Lastly, museum and libary are two activities which were not
attractive to this group and they did not vist in the past 12 months.
Cluster 1
This cluster primarily includes people in the 45-54 years old age group, with mostly males living in the city
Activities: On average, cooking and washing dishes take up the most time each day (16.05 and 5.3 minutes, respectively). The next most
time is spent on listenting to radio and phonecalls which accounts for about 4.6 and 4 minutes per day on average. Visiting libraries is a
common interest among this group whereas visiting museums has not been in their interests in the past 12 months.
Cluster 2
Cluster 0
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 14/36
This cluster primarily includes people in the 45-54 years old age group, with mostly females living in the city
Activities: On average, cooking and washing dishes take up the most time each day (80.3 and 29.3 minutes, respectively). The next most
time is spent on listenting to radio and phonecalls which accounts for 3.3 and 12.08 minutes per day on average. Visiting libraries is a
preferred activity for this group whereas visiting museums has not been in their interests in the past 12 months.
Summary
Cluster 0 and 1 have gender and living area in common (male and city). Cluster 0 tends to spend more time on listening to radio and
cooking whereas cluster 1 prefer visiting library and spend time on other categories than cluster 0. The cluster 2 has a similar age with
cluster 1 and similar habits to visit library in the past 12 months, and this cluster spend the most their time on cooking and washing
dishes as well as phone calls in comparison to the other 2 clusters. This is an indication that females in the sample group tended to
spend more time on housework than other activities whereas males preferred enjoying other activities.
We liked to see, how many individuals was there in each cluster to verify our ndings.
# Checking how many people per cluster
print(data[data.Cluster == 0].shape[0])
print(data[data.Cluster == 1].shape[0])
print(data[data.Cluster == 2].shape[0])
6
227
110
As we can see, these clusters have large differences in their populations, which can affect the results and values the groups get. A single outlier
in a small group can have a large impact on the results, as we see with the amount of time spent listening to radio by cluster 0, which only has 6
people in it.
We want to know more about the individuals in the data outside of clustering so we calculated and visualised how participants in different
demographics spend their time to have a better overiview of the individuals' behavior.
# Calculate the average time spent on differences in activities based on day of week
activity_means = data.groupby('Day of week')[numerical_columns].mean()
# Display the average time for activities by day of week
print("Average Time Spent on Activities by day of week:\n", activity_means)
# Visualize the comparison using a bar plot
activity_means.plot(kind='bar', figsize=(10, 6))
plt.title('Average Time Spent on Activities by day of week')
plt.ylabel('Average Time (minutes)')
plt.xlabel('Day of week')
plt.legend(title='Activity', loc='upper right')
plt.tight_layout()
plt.show()
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 15/36
Activities: On average, cooking and washing dishes take up the most time each day (80.3 and 29.3 minutes, respectively). The next most
time is spent on listenting to radio and phonecalls which accounts for 3.3 and 12.08 minutes per day on average. Visiting libraries is a
preferred activity for this group whereas visiting museums has not been in their interests in the past 12 months.
Summary
Cluster 0 and 1 have gender and living area in common (male and city). Cluster 0 tends to spend more time on listening to radio and
cooking whereas cluster 1 prefer visiting library and spend time on other categories than cluster 0. The cluster 2 has a similar age with
cluster 1 and similar habits to visit library in the past 12 months, and this cluster spend the most their time on cooking and washing
dishes as well as phone calls in comparison to the other 2 clusters. This is an indication that females in the sample group tended to
spend more time on housework than other activities whereas males preferred enjoying other activities.
We liked to see, how many individuals was there in each cluster to verify our ndings.
# Checking how many people per cluster
print(data[data.Cluster == 0].shape[0])
print(data[data.Cluster == 1].shape[0])
print(data[data.Cluster == 2].shape[0])
6
227
110
As we can see, these clusters have large differences in their populations, which can affect the results and values the groups get. A single outlier
in a small group can have a large impact on the results, as we see with the amount of time spent listening to radio by cluster 0, which only has 6
people in it.
We want to know more about the individuals in the data outside of clustering so we calculated and visualised how participants in different
demographics spend their time to have a better overiview of the individuals' behavior.
# Calculate the average time spent on differences in activities based on day of week
activity_means = data.groupby('Day of week')[numerical_columns].mean()
# Display the average time for activities by day of week
print("Average Time Spent on Activities by day of week:\n", activity_means)
# Visualize the comparison using a bar plot
activity_means.plot(kind='bar', figsize=(10, 6))
plt.title('Average Time Spent on Activities by day of week')
plt.ylabel('Average Time (minutes)')
plt.xlabel('Day of week')
plt.legend(title='Activity', loc='upper right')
plt.tight_layout()
plt.show()
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 15/36
Average Time Spent on Activities by day of week:
Cooking Washing dishes Listening to radio Phonecall
Day of week
weekend 33.595506 9.831461 9.213483 5.842697
working day 40.545455 15.696970 7.878788 7.030303
<ipython-input-27-c44c9dee0fab>:2: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
activity_means = data.groupby('Day of week')[numerical_columns].mean()
# Calculate the average time spent on differences in activities based on sex
activity_means = data.groupby('Sex')[numerical_columns].mean()
# Display the average time for activities by sex
print("Average Time Spent on Activities by sex:\n", activity_means)
# Visualize the comparison using a bar plot
activity_means.plot(kind='bar', figsize=(10, 6))
plt.title('Average Time Spent on Activities by sex')
plt.ylabel('Average Time (minutes)')
plt.xlabel('Sex')
plt.legend(title='Activity', loc='upper right')
plt.tight_layout()
plt.show()
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 16/36
Cooking Washing dishes Listening to radio Phonecall
Day of week
weekend 33.595506 9.831461 9.213483 5.842697
working day 40.545455 15.696970 7.878788 7.030303
<ipython-input-27-c44c9dee0fab>:2: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
activity_means = data.groupby('Day of week')[numerical_columns].mean()
# Calculate the average time spent on differences in activities based on sex
activity_means = data.groupby('Sex')[numerical_columns].mean()
# Display the average time for activities by sex
print("Average Time Spent on Activities by sex:\n", activity_means)
# Visualize the comparison using a bar plot
activity_means.plot(kind='bar', figsize=(10, 6))
plt.title('Average Time Spent on Activities by sex')
plt.ylabel('Average Time (minutes)')
plt.xlabel('Sex')
plt.legend(title='Activity', loc='upper right')
plt.tight_layout()
plt.show()
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 16/36
<ipython-input-28-ffcca95a3741>:2: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
activity_means = data.groupby('Sex')[numerical_columns].mean()
Average Time Spent on Activities by sex:
Cooking Washing dishes Listening to radio Phonecall
Sex
Female 45.000000 17.105263 3.473684 9.052632
Male 26.928105 7.124183 14.901961 3.137255
# Calculate the average time spent on differences in activities based on age group
activity_means = data.groupby('Age group')[numerical_columns].mean()
# Display the average time for activities by age group
print("Average Time Spent on Activities by age group:\n", activity_means)
# Visualize the comparison using a bar plot
activity_means.plot(kind='bar', figsize=(10, 6))
plt.title('Average Time Spent on Activities by age group')
plt.ylabel('Average Time (minutes)')
plt.xlabel('Age group')
plt.legend(title='Activity', loc='upper left')
plt.tight_layout()
plt.show()
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 17/36
activity_means = data.groupby('Sex')[numerical_columns].mean()
Average Time Spent on Activities by sex:
Cooking Washing dishes Listening to radio Phonecall
Sex
Female 45.000000 17.105263 3.473684 9.052632
Male 26.928105 7.124183 14.901961 3.137255
# Calculate the average time spent on differences in activities based on age group
activity_means = data.groupby('Age group')[numerical_columns].mean()
# Display the average time for activities by age group
print("Average Time Spent on Activities by age group:\n", activity_means)
# Visualize the comparison using a bar plot
activity_means.plot(kind='bar', figsize=(10, 6))
plt.title('Average Time Spent on Activities by age group')
plt.ylabel('Average Time (minutes)')
plt.xlabel('Age group')
plt.legend(title='Activity', loc='upper left')
plt.tight_layout()
plt.show()
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 17/36
<ipython-input-29-464a028e00ca>:2: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
activity_means = data.groupby('Age group')[numerical_columns].mean()
Average Time Spent on Activities by age group:
Cooking Washing dishes Listening to radio Phonecall
Age group
20-24 years old 12.727273 2.727273 0.000000 0.909091
25-34 years old 27.291667 7.500000 0.000000 4.583333
35-44 years old 25.000000 10.192308 2.115385 4.807692
45-54 years old 31.000000 11.750000 9.875000 5.625000
55-64 years old 36.125000 9.750000 8.125000 5.875000
65-74 years old 55.531915 16.595745 15.957447 7.659574
75+ years old 77.600000 36.800000 25.600000 17.600000
# Calculate the average time spent on differences in activities based on living environments
activity_means = data.groupby('Living environment')[numerical_columns].mean()
# Display the average time for activities by living environment
print("Average Time Spent on Activities by living Environment:\n", activity_means)
# Visualize the comparison using a bar plot
activity_means.plot(kind='bar', figsize=(10, 6))
plt.title('Average Time Spent on Activities by living Environment')
plt.ylabel('Average Time (minutes)')
plt.xlabel('Living Environment')
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 18/36
activity_means = data.groupby('Age group')[numerical_columns].mean()
Average Time Spent on Activities by age group:
Cooking Washing dishes Listening to radio Phonecall
Age group
20-24 years old 12.727273 2.727273 0.000000 0.909091
25-34 years old 27.291667 7.500000 0.000000 4.583333
35-44 years old 25.000000 10.192308 2.115385 4.807692
45-54 years old 31.000000 11.750000 9.875000 5.625000
55-64 years old 36.125000 9.750000 8.125000 5.875000
65-74 years old 55.531915 16.595745 15.957447 7.659574
75+ years old 77.600000 36.800000 25.600000 17.600000
# Calculate the average time spent on differences in activities based on living environments
activity_means = data.groupby('Living environment')[numerical_columns].mean()
# Display the average time for activities by living environment
print("Average Time Spent on Activities by living Environment:\n", activity_means)
# Visualize the comparison using a bar plot
activity_means.plot(kind='bar', figsize=(10, 6))
plt.title('Average Time Spent on Activities by living Environment')
plt.ylabel('Average Time (minutes)')
plt.xlabel('Living Environment')
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 18/36
<ipython-input-30-ba1203361333>:2: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
activity_means = data.groupby('Living environment')[numerical_columns].mean()
Average Time Spent on Activities by living Environment:
Cooking Washing dishes Listening to radio Phonecall
Living environment
City 35.047170 11.273585 10.990566 7.075472
Municipality 38.909091 12.545455 0.909091 4.909091
Rural 40.789474 16.578947 7.368421 5.657895
plt.legend(title='Activity', loc='upper right')
plt.tight_layout()
plt.show()
Overall the most time is spent on cooking and washing dishes. Older people in the population seem to have more time for activities on average
based on these graphs. On average, people living in rural areas spent more time on these activities than others. People living in cities spent the
second most time on them and people in municipalities the least. Women spent almost 50% more time cooking and washing dishes on average
than men.
We'll like to know more about the individuals in the data, so we analysed and visualized frequency of people visiting museums and libraries
compared to age group, sex and living environment to have a better overiview of the individuals' behavior.
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 19/36
activity_means = data.groupby('Living environment')[numerical_columns].mean()
Average Time Spent on Activities by living Environment:
Cooking Washing dishes Listening to radio Phonecall
Living environment
City 35.047170 11.273585 10.990566 7.075472
Municipality 38.909091 12.545455 0.909091 4.909091
Rural 40.789474 16.578947 7.368421 5.657895
plt.legend(title='Activity', loc='upper right')
plt.tight_layout()
plt.show()
Overall the most time is spent on cooking and washing dishes. Older people in the population seem to have more time for activities on average
based on these graphs. On average, people living in rural areas spent more time on these activities than others. People living in cities spent the
second most time on them and people in municipalities the least. Women spent almost 50% more time cooking and washing dishes on average
than men.
We'll like to know more about the individuals in the data, so we analysed and visualized frequency of people visiting museums and libraries
compared to age group, sex and living environment to have a better overiview of the individuals' behavior.
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 19/36
<ipython-input-31-ca51556be1fb>:2: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
counts = data.groupby(['Sex','Museum']).size().unstack()
<ipython-input-31-ca51556be1fb>:3: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
data.groupby(['Sex','Museum']).size().unstack()
Museum No Yes
Sex
Female 106 84
Male 83 70
# Compare sex to visiting a museum in the past year
counts = data.groupby(['Sex','Museum']).size().unstack()
data.groupby(['Sex','Museum']).size().unstack()
<Axes: title={'center': 'How many female and male have visited the museum in the past year'}, xlabel='Sex', ylabel='Relative frequency'>
# Since there are different size for both group, relative frequencies should be calculated
# Totals by value of 'sex' (i.e. sums over columns)
totals = counts.sum(axis=1)
# Relative frequencies per value of 'sex'
# (i.e. divide each element by the total that corresponds to its row name)
relative_counts = counts.div(totals, axis=0)
relative_counts.plot.bar(title='How many female and male have visited the museum in the past year', ylabel='Relative frequency')
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 20/36
counts = data.groupby(['Sex','Museum']).size().unstack()
<ipython-input-31-ca51556be1fb>:3: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
data.groupby(['Sex','Museum']).size().unstack()
Museum No Yes
Sex
Female 106 84
Male 83 70
# Compare sex to visiting a museum in the past year
counts = data.groupby(['Sex','Museum']).size().unstack()
data.groupby(['Sex','Museum']).size().unstack()
<Axes: title={'center': 'How many female and male have visited the museum in the past year'}, xlabel='Sex', ylabel='Relative frequency'>
# Since there are different size for both group, relative frequencies should be calculated
# Totals by value of 'sex' (i.e. sums over columns)
totals = counts.sum(axis=1)
# Relative frequencies per value of 'sex'
# (i.e. divide each element by the total that corresponds to its row name)
relative_counts = counts.div(totals, axis=0)
relative_counts.plot.bar(title='How many female and male have visited the museum in the past year', ylabel='Relative frequency')
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 20/36
<ipython-input-33-25dabdfed02c>:2: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
counts = data.groupby(['Sex','Library']).size().unstack()
<ipython-input-33-25dabdfed02c>:3: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
data.groupby(['Sex','Library']).size().unstack()
Library No Yes
Sex
Female 52 138
Male 60 93
# Compare sex to visiting a library in the past year
counts = data.groupby(['Sex','Library']).size().unstack()
data.groupby(['Sex','Library']).size().unstack()
<Axes: title={'center': 'How many female and male have visited the library in the past year'}, xlabel='Sex', ylabel='Relative frequency'>
totals = counts.sum(axis=1)
relative_counts = counts.div(totals, axis=0)
relative_counts.plot.bar(title='How many female and male have visited the library in the past year', ylabel='Relative frequency')
# Compare age groups to visiting a museum in the past year
counts = data.groupby(['Age group','Museum']).size().unstack()
data.groupby(['Age group','Museum']).size().unstack()
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 21/36
counts = data.groupby(['Sex','Library']).size().unstack()
<ipython-input-33-25dabdfed02c>:3: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
data.groupby(['Sex','Library']).size().unstack()
Library No Yes
Sex
Female 52 138
Male 60 93
# Compare sex to visiting a library in the past year
counts = data.groupby(['Sex','Library']).size().unstack()
data.groupby(['Sex','Library']).size().unstack()
<Axes: title={'center': 'How many female and male have visited the library in the past year'}, xlabel='Sex', ylabel='Relative frequency'>
totals = counts.sum(axis=1)
relative_counts = counts.div(totals, axis=0)
relative_counts.plot.bar(title='How many female and male have visited the library in the past year', ylabel='Relative frequency')
# Compare age groups to visiting a museum in the past year
counts = data.groupby(['Age group','Museum']).size().unstack()
data.groupby(['Age group','Museum']).size().unstack()
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 21/36
<ipython-input-35-bcf623d84b13>:2: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
counts = data.groupby(['Age group','Museum']).size().unstack()
<ipython-input-35-bcf623d84b13>:3: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
data.groupby(['Age group','Museum']).size().unstack()
Museum No Yes
Age group
20-24 years old 7 4
25-34 years old 30 18
35-44 years old 23 29
45-54 years old 39 41
55-64 years old 42 38
65-74 years old 30 17
75+ years old 18 7
<Axes: title={'center': 'How many have visited the museum in the past year within an age group'}, xlabel='Age group', ylabel='Relative frequency'>
totals = counts.sum(axis=1)
relative_counts = counts.div(totals, axis=0)
relative_counts.plot.bar(title='How many have visited the museum in the past year within an age group', ylabel='Relative frequency')
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 22/36
counts = data.groupby(['Age group','Museum']).size().unstack()
<ipython-input-35-bcf623d84b13>:3: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
data.groupby(['Age group','Museum']).size().unstack()
Museum No Yes
Age group
20-24 years old 7 4
25-34 years old 30 18
35-44 years old 23 29
45-54 years old 39 41
55-64 years old 42 38
65-74 years old 30 17
75+ years old 18 7
<Axes: title={'center': 'How many have visited the museum in the past year within an age group'}, xlabel='Age group', ylabel='Relative frequency'>
totals = counts.sum(axis=1)
relative_counts = counts.div(totals, axis=0)
relative_counts.plot.bar(title='How many have visited the museum in the past year within an age group', ylabel='Relative frequency')
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 22/36
<ipython-input-37-0fc4cbdcb4e3>:2: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
counts = data.groupby(['Age group','Library']).size().unstack()
<ipython-input-37-0fc4cbdcb4e3>:3: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
data.groupby(['Age group','Library']).size().unstack()
Library No Yes
Age group
20-24 years old 0 11
25-34 years old 3 45
35-44 years old 11 41
45-54 years old 27 53
55-64 years old 33 47
65-74 years old 23 24
75+ years old 15 10
# Compare age groups to visiting a library in the past year
counts = data.groupby(['Age group','Library']).size().unstack()
data.groupby(['Age group','Library']).size().unstack()
totals = counts.sum(axis=1)
relative_counts = counts.div(totals, axis=0)
relative_counts.plot.bar(title='How many have visited the library in the past year within an age group', ylabel='Relative frequency')
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 23/36
counts = data.groupby(['Age group','Library']).size().unstack()
<ipython-input-37-0fc4cbdcb4e3>:3: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
data.groupby(['Age group','Library']).size().unstack()
Library No Yes
Age group
20-24 years old 0 11
25-34 years old 3 45
35-44 years old 11 41
45-54 years old 27 53
55-64 years old 33 47
65-74 years old 23 24
75+ years old 15 10
# Compare age groups to visiting a library in the past year
counts = data.groupby(['Age group','Library']).size().unstack()
data.groupby(['Age group','Library']).size().unstack()
totals = counts.sum(axis=1)
relative_counts = counts.div(totals, axis=0)
relative_counts.plot.bar(title='How many have visited the library in the past year within an age group', ylabel='Relative frequency')
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 23/36
<Axes: title={'center': 'How many have visited the library in the past year within an age group'}, xlabel='Age group', ylabel='Relative frequency'>
<ipython-input-39-c552f616ea2e>:2: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
counts = data.groupby(['Living environment','Museum']).size().unstack()
<ipython-input-39-c552f616ea2e>:3: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
data.groupby(['Living environment','Museum']).size().unstack()
Museum No Yes
Living environment
City 114 98
Municipality 29 26
Rural 46 30
# Compare living environment to visiting a museum in the past year
counts = data.groupby(['Living environment','Museum']).size().unstack()
data.groupby(['Living environment','Museum']).size().unstack()
totals = counts.sum(axis=1)
relative_counts = counts.div(totals, axis=0)
relative_counts.plot.bar(title='How many have visited the museum in the past year in each living environment', ylabel='Relative frequency')
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 24/36
<ipython-input-39-c552f616ea2e>:2: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
counts = data.groupby(['Living environment','Museum']).size().unstack()
<ipython-input-39-c552f616ea2e>:3: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
data.groupby(['Living environment','Museum']).size().unstack()
Museum No Yes
Living environment
City 114 98
Municipality 29 26
Rural 46 30
# Compare living environment to visiting a museum in the past year
counts = data.groupby(['Living environment','Museum']).size().unstack()
data.groupby(['Living environment','Museum']).size().unstack()
totals = counts.sum(axis=1)
relative_counts = counts.div(totals, axis=0)
relative_counts.plot.bar(title='How many have visited the museum in the past year in each living environment', ylabel='Relative frequency')
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 24/36
<Axes: title={'center': 'How many have visited the museum in the past year in each living environment'}, xlabel='Living environment', ylabel='Relative frequency'>
<ipython-input-41-3c4011b7d4fe>:2: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
counts = data.groupby(['Living environment','Library']).size().unstack()
<ipython-input-41-3c4011b7d4fe>:3: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
data.groupby(['Living environment','Library']).size().unstack()
Library No Yes
Living environment
City 70 142
Municipality 16 39
Rural 26 50
# Compare living environment to visiting a library in the past year
counts = data.groupby(['Living environment','Library']).size().unstack()
data.groupby(['Living environment','Library']).size().unstack()
totals = counts.sum(axis=1)
relative_counts = counts.div(totals, axis=0)
relative_counts.plot.bar(title='How many have visited the library in the past year in each living environment', ylabel='Relative frequency')
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 25/36
<ipython-input-41-3c4011b7d4fe>:2: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
counts = data.groupby(['Living environment','Library']).size().unstack()
<ipython-input-41-3c4011b7d4fe>:3: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
data.groupby(['Living environment','Library']).size().unstack()
Library No Yes
Living environment
City 70 142
Municipality 16 39
Rural 26 50
# Compare living environment to visiting a library in the past year
counts = data.groupby(['Living environment','Library']).size().unstack()
data.groupby(['Living environment','Library']).size().unstack()
totals = counts.sum(axis=1)
relative_counts = counts.div(totals, axis=0)
relative_counts.plot.bar(title='How many have visited the library in the past year in each living environment', ylabel='Relative frequency')
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 25/36
<Axes: title={'center': 'How many have visited the library in the past year in each living environment'}, xlabel='Living environment', ylabel='Relative frequency'>
Based on these graphs, while there was not much difference between males and females in visiting libraries or museums, libraries were visited
signi cantly more than museums. Younger and older people in this study's population seem to visit museums less than middle-aged people, of
which almost half have visited in the past year. Library visits however seem steadily to lessen as the person gets older.
2. Estimating how much time is spent on each activity on average
We can assume that the sample, which our estimation is based on, is a representative of the population.
We will calculate and visualise average time spent on activities by household.
# estimate how much household spends time on average daily by activity
mean_of_activies_by_household = data.groupby('household ID')[numerical_columns].sum().mean()
data.groupby('household ID')[numerical_columns].sum().mean()
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 26/36
Based on these graphs, while there was not much difference between males and females in visiting libraries or museums, libraries were visited
signi cantly more than museums. Younger and older people in this study's population seem to visit museums less than middle-aged people, of
which almost half have visited in the past year. Library visits however seem steadily to lessen as the person gets older.
2. Estimating how much time is spent on each activity on average
We can assume that the sample, which our estimation is based on, is a representative of the population.
We will calculate and visualise average time spent on activities by household.
# estimate how much household spends time on average daily by activity
mean_of_activies_by_household = data.groupby('household ID')[numerical_columns].sum().mean()
data.groupby('household ID')[numerical_columns].sum().mean()
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 26/36
0
Cooking 36.938776
Washing dishes 12.653061
Listening to radio 8.571429
Phonecall 6.413994
dtype: float64
<Axes: title={'center': 'Average of time spend on activities by households'}, ylabel='Minutes'>
mean_of_activies_by_household.plot.bar(title='Average of time spend on activities by households',
ylabel='Minutes')
Based on the estimation it seems that the households in the population tend to use the most time on cooking and the least on talking on the
phone.
3. Finding differences in time spent on the activities by day of week
Firstly we calculate and visualise the average time spent on activities based on the day of week.
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 27/36
Cooking 36.938776
Washing dishes 12.653061
Listening to radio 8.571429
Phonecall 6.413994
dtype: float64
<Axes: title={'center': 'Average of time spend on activities by households'}, ylabel='Minutes'>
mean_of_activies_by_household.plot.bar(title='Average of time spend on activities by households',
ylabel='Minutes')
Based on the estimation it seems that the households in the population tend to use the most time on cooking and the least on talking on the
phone.
3. Finding differences in time spent on the activities by day of week
Firstly we calculate and visualise the average time spent on activities based on the day of week.
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 27/36
<ipython-input-45-32d12b607e76>:4: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
activity_means = data.groupby('Day of week')[numerical_columns].mean()
Average Time Spent on Activities by day of week:
Cooking Washing dishes Listening to radio Phonecall
Day of week
weekend 33.595506 9.831461 9.213483 5.842697
working day 40.545455 15.696970 7.878788 7.030303
##This is a duplicate from section 1. where we used it to characterise individuals
# Calculate the average time spent on differences in activities based on day of week
activity_means = data.groupby('Day of week')[numerical_columns].mean()
# Display the average time for activities by day of week
print("Average Time Spent on Activities by day of week:\n", activity_means)
# Visualize the comparison using a bar plot
activity_means.plot(kind='bar', figsize=(10, 6))
plt.title('Average Time Spent on Activities by day of week')
plt.ylabel('Average Time (minutes)')
plt.xlabel('Day of week')
plt.legend(title='Activity', loc='upper right')
plt.tight_layout()
plt.show()
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 28/36
activity_means = data.groupby('Day of week')[numerical_columns].mean()
Average Time Spent on Activities by day of week:
Cooking Washing dishes Listening to radio Phonecall
Day of week
weekend 33.595506 9.831461 9.213483 5.842697
working day 40.545455 15.696970 7.878788 7.030303
##This is a duplicate from section 1. where we used it to characterise individuals
# Calculate the average time spent on differences in activities based on day of week
activity_means = data.groupby('Day of week')[numerical_columns].mean()
# Display the average time for activities by day of week
print("Average Time Spent on Activities by day of week:\n", activity_means)
# Visualize the comparison using a bar plot
activity_means.plot(kind='bar', figsize=(10, 6))
plt.title('Average Time Spent on Activities by day of week')
plt.ylabel('Average Time (minutes)')
plt.xlabel('Day of week')
plt.legend(title='Activity', loc='upper right')
plt.tight_layout()
plt.show()
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 28/36
By looking at the graph above, we can have an overall view about time spent activities between working days and weekends:
Cooking: time spent on cooking mostly on working days (40.54 minutes) which is higher than weekends (33.59 minutes). It means that
people prefer less cooking on weekends so that they can rest or spend time on other activities
Washing dishes: Time spent on washing dishes (15.69 minutes) is higher than weekends (9.8 minutes), which indicates people prefer
eating at home on working days, leading to more time washing dishes, while on weekends they may like to eat outside and less cooking.
Listening to Radio: There is no big difference in radio listening time between workdays (7.87 minutes) and weekends (9.21 minutes). This
shows that the habits of listening to radio do not change much between workdays and weekends, they may listen in their free time or
when doing other things.
Phone call: The average time spent making phone calls is slightly higher on workdays (7.03 minutes) than on weekends (5.84 minutes).
The reason can be work-related calls is more than on weekends.
Overall, time spent on cooking and dishwashing activities seemed to decrease on weekends, while time spent on activities such as listening to
the radio, making phone calls, visiting museums, and visiting the library did not change too much between workdays and weekends. Heavy
housework activities such as cooking and washing dishes tend to decrease on weekends, possibly re ecting the need to rest and reduce stress
on these days.
Summary
Next we will apply statistical testing to nd evidence if day of the week is associated to time spent on activities.
First step is to test if the data is normal. We'll test the variables' normality with Shapiro-Wilk test. The null hypothesis is that is that the dataset
is normally distributed [1].
# Test for normality
normality = lambda a: s.shapiro(a).pvalue
print(data.groupby('Day of week')['Cooking'].apply(normality))
print(data.groupby('Day of week')['Washing dishes'].apply(normality))
print(data.groupby('Day of week')['Listening to radio'].apply(normality))
print(data.groupby('Day of week')['Phonecall'].apply(normality))
Day of week
weekend 3.683398e-17
working day 1.805845e-11
Name: Cooking, dtype: float64
Day of week
weekend 5.364922e-19
working day 9.148754e-15
Name: Washing dishes, dtype: float64
Day of week
weekend 3.926256e-25
working day 1.654761e-26
Name: Listening to radio, dtype: float64
Day of week
weekend 1.249983e-23
working day 9.539748e-22
Name: Phonecall, dtype: float64
<ipython-input-46-36d2319ad690>:3: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
print(data.groupby('Day of week')['Cooking'].apply(normality))
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 29/36
Cooking: time spent on cooking mostly on working days (40.54 minutes) which is higher than weekends (33.59 minutes). It means that
people prefer less cooking on weekends so that they can rest or spend time on other activities
Washing dishes: Time spent on washing dishes (15.69 minutes) is higher than weekends (9.8 minutes), which indicates people prefer
eating at home on working days, leading to more time washing dishes, while on weekends they may like to eat outside and less cooking.
Listening to Radio: There is no big difference in radio listening time between workdays (7.87 minutes) and weekends (9.21 minutes). This
shows that the habits of listening to radio do not change much between workdays and weekends, they may listen in their free time or
when doing other things.
Phone call: The average time spent making phone calls is slightly higher on workdays (7.03 minutes) than on weekends (5.84 minutes).
The reason can be work-related calls is more than on weekends.
Overall, time spent on cooking and dishwashing activities seemed to decrease on weekends, while time spent on activities such as listening to
the radio, making phone calls, visiting museums, and visiting the library did not change too much between workdays and weekends. Heavy
housework activities such as cooking and washing dishes tend to decrease on weekends, possibly re ecting the need to rest and reduce stress
on these days.
Summary
Next we will apply statistical testing to nd evidence if day of the week is associated to time spent on activities.
First step is to test if the data is normal. We'll test the variables' normality with Shapiro-Wilk test. The null hypothesis is that is that the dataset
is normally distributed [1].
# Test for normality
normality = lambda a: s.shapiro(a).pvalue
print(data.groupby('Day of week')['Cooking'].apply(normality))
print(data.groupby('Day of week')['Washing dishes'].apply(normality))
print(data.groupby('Day of week')['Listening to radio'].apply(normality))
print(data.groupby('Day of week')['Phonecall'].apply(normality))
Day of week
weekend 3.683398e-17
working day 1.805845e-11
Name: Cooking, dtype: float64
Day of week
weekend 5.364922e-19
working day 9.148754e-15
Name: Washing dishes, dtype: float64
Day of week
weekend 3.926256e-25
working day 1.654761e-26
Name: Listening to radio, dtype: float64
Day of week
weekend 1.249983e-23
working day 9.539748e-22
Name: Phonecall, dtype: float64
<ipython-input-46-36d2319ad690>:3: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
print(data.groupby('Day of week')['Cooking'].apply(normality))
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 29/36
<ipython-input-46-36d2319ad690>:4: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
print(data.groupby('Day of week')['Washing dishes'].apply(normality))
<ipython-input-46-36d2319ad690>:5: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
print(data.groupby('Day of week')['Listening to radio'].apply(normality))
<ipython-input-46-36d2319ad690>:6: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
print(data.groupby('Day of week')['Phonecall'].apply(normality))
We can assume normality based on the high p-values obtained in the test for normality for all variables. Here we'll test two unpaired variables
so we will apply unpaired T-test. The test's null hypothesis states that there is no signi cant difference between the means of the two groups
[2].
# assume normality due to high p-values, test with unpaired T-test
p_t_cooking = s.ttest_ind(data[data['Day of week'] == 'working day']['Cooking'], data[data['Day of week'] == 'weekend']['Cooking']).pvalue
p_t_dishes = s.ttest_ind(data[data['Day of week'] == 'working day']['Washing dishes'], data[data['Day of week'] == 'weekend']['Washing dishes']).pvalue
p_t_radio = s.ttest_ind(data[data['Day of week'] == 'working day']['Listening to radio'], data[data['Day of week'] == 'weekend']['Listening to radio']).pvalue
p_t_phonecall = s.ttest_ind(data[data['Day of week'] == 'working day']['Phonecall'], data[data['Day of week'] == 'weekend']['Phonecall']).pvalue
print(p_t_cooking)
print(p_t_dishes)
print(p_t_radio)
print(p_t_phonecall)
0.11924730669038067
0.004017561018094432
0.7420208251463922
0.5038410577452723
print(data[data['Day of week'] == 'working day']['Washing dishes'].mean())
print(data[data['Day of week'] == 'weekend']['Washing dishes'].mean())
15.696969696969697
9.831460674157304
Based on the p-values obtained by unpaired T-test, we can see that there's no statistical evidence that time spent on cooking, listening to radio
or phonecall have difference between working days and weekend. However, there's evidence that washing dishes tend to be done more on one
group than the other.
Since we are running multiple tests in the same dataset simultaneously, we should correct the tests' p-values. This is done to identify if the false
positivies in the test results [3]. We will use Bonferroni correction to correct the p-values.
# Use Bonferroni correction
p_values = [p_t_cooking, p_t_dishes, p_t_radio, p_t_phonecall]
ssm.multipletests(p_values, method='bonferroni')[1]
array([0.47698923, 0.01607024, 1. , 1. ])
After correcting the values we can see that there's no difference in the signi cance, so the evidence found before seems to be valid.
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 30/36
print(data.groupby('Day of week')['Washing dishes'].apply(normality))
<ipython-input-46-36d2319ad690>:5: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
print(data.groupby('Day of week')['Listening to radio'].apply(normality))
<ipython-input-46-36d2319ad690>:6: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
print(data.groupby('Day of week')['Phonecall'].apply(normality))
We can assume normality based on the high p-values obtained in the test for normality for all variables. Here we'll test two unpaired variables
so we will apply unpaired T-test. The test's null hypothesis states that there is no signi cant difference between the means of the two groups
[2].
# assume normality due to high p-values, test with unpaired T-test
p_t_cooking = s.ttest_ind(data[data['Day of week'] == 'working day']['Cooking'], data[data['Day of week'] == 'weekend']['Cooking']).pvalue
p_t_dishes = s.ttest_ind(data[data['Day of week'] == 'working day']['Washing dishes'], data[data['Day of week'] == 'weekend']['Washing dishes']).pvalue
p_t_radio = s.ttest_ind(data[data['Day of week'] == 'working day']['Listening to radio'], data[data['Day of week'] == 'weekend']['Listening to radio']).pvalue
p_t_phonecall = s.ttest_ind(data[data['Day of week'] == 'working day']['Phonecall'], data[data['Day of week'] == 'weekend']['Phonecall']).pvalue
print(p_t_cooking)
print(p_t_dishes)
print(p_t_radio)
print(p_t_phonecall)
0.11924730669038067
0.004017561018094432
0.7420208251463922
0.5038410577452723
print(data[data['Day of week'] == 'working day']['Washing dishes'].mean())
print(data[data['Day of week'] == 'weekend']['Washing dishes'].mean())
15.696969696969697
9.831460674157304
Based on the p-values obtained by unpaired T-test, we can see that there's no statistical evidence that time spent on cooking, listening to radio
or phonecall have difference between working days and weekend. However, there's evidence that washing dishes tend to be done more on one
group than the other.
Since we are running multiple tests in the same dataset simultaneously, we should correct the tests' p-values. This is done to identify if the false
positivies in the test results [3]. We will use Bonferroni correction to correct the p-values.
# Use Bonferroni correction
p_values = [p_t_cooking, p_t_dishes, p_t_radio, p_t_phonecall]
ssm.multipletests(p_values, method='bonferroni')[1]
array([0.47698923, 0.01607024, 1. , 1. ])
After correcting the values we can see that there's no difference in the signi cance, so the evidence found before seems to be valid.
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 30/36
4. Finding differences in time spent on the activities by living environment
Firstly we calculate and visualise the average time spent on activities based on living environments.
##This is a duplicate from section 1. where we used it to characterise individuals
# Calculate the average time spent on differences in activities based on living environments
activity_means = data.groupby('Living environment')[numerical_columns].mean()
# Display the average time for activities by living environment
print("Average Time Spent on Activities by living Environment:\n", activity_means)
# Visualize the comparison using a bar plot
activity_means.plot(kind='bar', figsize=(10, 6))
plt.title('Average Time Spent on Activities by living Environment')
plt.ylabel('Average Time (minutes)')
plt.xlabel('Living Environment')
plt.legend(title='Activity', loc='upper right')
plt.tight_layout()
plt.show()
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 31/36
Firstly we calculate and visualise the average time spent on activities based on living environments.
##This is a duplicate from section 1. where we used it to characterise individuals
# Calculate the average time spent on differences in activities based on living environments
activity_means = data.groupby('Living environment')[numerical_columns].mean()
# Display the average time for activities by living environment
print("Average Time Spent on Activities by living Environment:\n", activity_means)
# Visualize the comparison using a bar plot
activity_means.plot(kind='bar', figsize=(10, 6))
plt.title('Average Time Spent on Activities by living Environment')
plt.ylabel('Average Time (minutes)')
plt.xlabel('Living Environment')
plt.legend(title='Activity', loc='upper right')
plt.tight_layout()
plt.show()
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 31/36
<ipython-input-50-4e2541126dcc>:4: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
activity_means = data.groupby('Living environment')[numerical_columns].mean()
Average Time Spent on Activities by living Environment:
Cooking Washing dishes Listening to radio Phonecall
Living environment
City 35.047170 11.273585 10.990566 7.075472
Municipality 38.909091 12.545455 0.909091 4.909091
Rural 40.789474 16.578947 7.368421 5.657895
In the average time spent on activities based on living environment, cooking and washing dishes account for the most time across all areas.
The rural area spends the most time on both cooking and washing dishes compared to city and municipality. Additionally, the residents in city
and rural tends to spend more time on listening to radio than those in municipality. Hoever, the time spent on visiting museums and libraries is
quite similar across all three areas.
All regions tends to spend their times on cooking and washing dishes, with the city spent the least time on these activities.
The residents in the cityspent more time on listening to radio which is oppsite of those live in municipality area.
Time spent on visiting museum and library is quite similar across all living environment.
Summary
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 32/36
activity_means = data.groupby('Living environment')[numerical_columns].mean()
Average Time Spent on Activities by living Environment:
Cooking Washing dishes Listening to radio Phonecall
Living environment
City 35.047170 11.273585 10.990566 7.075472
Municipality 38.909091 12.545455 0.909091 4.909091
Rural 40.789474 16.578947 7.368421 5.657895
In the average time spent on activities based on living environment, cooking and washing dishes account for the most time across all areas.
The rural area spends the most time on both cooking and washing dishes compared to city and municipality. Additionally, the residents in city
and rural tends to spend more time on listening to radio than those in municipality. Hoever, the time spent on visiting museums and libraries is
quite similar across all three areas.
All regions tends to spend their times on cooking and washing dishes, with the city spent the least time on these activities.
The residents in the cityspent more time on listening to radio which is oppsite of those live in municipality area.
Time spent on visiting museum and library is quite similar across all living environment.
Summary
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 32/36
Next we'll try to nd statistical evidence of living environment associating with time spent on each activity by applying statistical test. Again the
rst step is to calculate normality, and we'll use Shapiro-Wilk test for that.
# Test for normality
normality = lambda a: s.shapiro(a).pvalue
print(data.groupby('Living environment')['Cooking'].apply(normality))
print(data.groupby('Living environment')['Washing dishes'].apply(normality))
print(data.groupby('Living environment')['Listening to radio'].apply(normality))
print(data.groupby('Living environment')['Phonecall'].apply(normality))
Living environment
City 1.295625e-17
Municipality 1.905467e-07
Rural 4.519422e-07
Name: Cooking, dtype: float64
Living environment
City 2.789426e-19
Municipality 2.988767e-07
Rural 1.448247e-10
Name: Washing dishes, dtype: float64
Living environment
City 2.781124e-28
Municipality 2.257244e-15
Rural 3.323618e-17
Name: Listening to radio, dtype: float64
Living environment
City 7.110511e-25
Municipality 2.876741e-13
Rural 3.085402e-15
Name: Phonecall, dtype: float64
<ipython-input-51-72f837b37030>:3: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
print(data.groupby('Living environment')['Cooking'].apply(normality))
<ipython-input-51-72f837b37030>:4: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
print(data.groupby('Living environment')['Washing dishes'].apply(normality))
<ipython-input-51-72f837b37030>:5: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
print(data.groupby('Living environment')['Listening to radio'].apply(normality))
<ipython-input-51-72f837b37030>:6: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
print(data.groupby('Living environment')['Phonecall'].apply(normality))
Due to the high p-values, we can assume that the data is normal. Since the data is normally distributed and we'll have to apply test for multiple
unpaired values, we will use ANOVA. The null hypothesis in ANOVA is that the means of the groups are not signi cantly different [5].
p_a_cooking = s.f_oneway(data[data['Living environment'] == 'City']['Cooking'],
data[data['Living environment'] == 'Municipality']['Cooking'],
data[data['Living environment'] == 'Rural']['Cooking'])
p_a_dishes = s.f_oneway(data[data['Living environment'] == 'City']['Washing dishes'],
data[data['Living environment'] == 'Municipality']['Washing dishes'],
data[data['Living environment'] == 'Rural']['Washing dishes'])
p_a_radio = s.f_oneway(data[data['Living environment'] == 'City']['Listening to radio'],
data[data['Living environment'] == 'Municipality']['Listening to radio'],
data[data['Living environment'] == 'Rural']['Listening to radio'])
p_a_phonecall = s.f_oneway(data[data['Living environment'] == 'City']['Phonecall'],
data[data['Living environment'] == 'Municipality']['Phonecall'],
data[data['Living environment'] == 'Rural']['Phonecall'])
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 33/36
rst step is to calculate normality, and we'll use Shapiro-Wilk test for that.
# Test for normality
normality = lambda a: s.shapiro(a).pvalue
print(data.groupby('Living environment')['Cooking'].apply(normality))
print(data.groupby('Living environment')['Washing dishes'].apply(normality))
print(data.groupby('Living environment')['Listening to radio'].apply(normality))
print(data.groupby('Living environment')['Phonecall'].apply(normality))
Living environment
City 1.295625e-17
Municipality 1.905467e-07
Rural 4.519422e-07
Name: Cooking, dtype: float64
Living environment
City 2.789426e-19
Municipality 2.988767e-07
Rural 1.448247e-10
Name: Washing dishes, dtype: float64
Living environment
City 2.781124e-28
Municipality 2.257244e-15
Rural 3.323618e-17
Name: Listening to radio, dtype: float64
Living environment
City 7.110511e-25
Municipality 2.876741e-13
Rural 3.085402e-15
Name: Phonecall, dtype: float64
<ipython-input-51-72f837b37030>:3: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
print(data.groupby('Living environment')['Cooking'].apply(normality))
<ipython-input-51-72f837b37030>:4: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
print(data.groupby('Living environment')['Washing dishes'].apply(normality))
<ipython-input-51-72f837b37030>:5: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
print(data.groupby('Living environment')['Listening to radio'].apply(normality))
<ipython-input-51-72f837b37030>:6: FutureWarning: The default of observed=False is deprecated and will be changed to True in a future version of pandas. Pass observed=False to retain current behav
print(data.groupby('Living environment')['Phonecall'].apply(normality))
Due to the high p-values, we can assume that the data is normal. Since the data is normally distributed and we'll have to apply test for multiple
unpaired values, we will use ANOVA. The null hypothesis in ANOVA is that the means of the groups are not signi cantly different [5].
p_a_cooking = s.f_oneway(data[data['Living environment'] == 'City']['Cooking'],
data[data['Living environment'] == 'Municipality']['Cooking'],
data[data['Living environment'] == 'Rural']['Cooking'])
p_a_dishes = s.f_oneway(data[data['Living environment'] == 'City']['Washing dishes'],
data[data['Living environment'] == 'Municipality']['Washing dishes'],
data[data['Living environment'] == 'Rural']['Washing dishes'])
p_a_radio = s.f_oneway(data[data['Living environment'] == 'City']['Listening to radio'],
data[data['Living environment'] == 'Municipality']['Listening to radio'],
data[data['Living environment'] == 'Rural']['Listening to radio'])
p_a_phonecall = s.f_oneway(data[data['Living environment'] == 'City']['Phonecall'],
data[data['Living environment'] == 'Municipality']['Phonecall'],
data[data['Living environment'] == 'Rural']['Phonecall'])
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 33/36
print(p_a_cooking)
print(p_a_dishes)
print(p_a_radio)
print(p_a_phonecall)
F_onewayResult(statistic=0.6150287800038885, pvalue=0.5412256844691998)
F_onewayResult(statistic=2.211383605174631, pvalue=0.11112227149093898)
F_onewayResult(statistic=1.63970308504325, pvalue=0.19556825223405488)
F_onewayResult(statistic=0.4827571963186801, pvalue=0.6175019597524363)
Based on the p-values obtained, there appears to be no statistical evidence that time spent on the activities has difference between the living
environments.
Since there's multiple simultaneous tests, we should apply multiple correction again.
# Use Bonferroni correction
p_values = [p_a_cooking[1], p_a_dishes[1], p_a_radio[1], p_a_phonecall[1]]
ssm.multipletests(p_values, method='bonferroni')[1]
array([1. , 0.44448909, 0.78227301, 1. ])
After correcting the p-values, we can see that there's no signi cant difference when comparing to the p-values obtained from the test.
5. Exploring which of the activities are associated with each other
Next we use another way of estimating and visualising correlations between activities using matrixes. First by using pearson correlation
coe cient and then by using spearman's rank correlation coe cient.
# Identify pearson correlation between activities
activity_corr = data[numerical_columns].corr()
print("Activity Correlation Matrix:\n", activity_corr)
# Visualize the correlation matrix
sns.heatmap(activity_corr, annot=True, cmap='coolwarm')
plt.title('Correlation Between Activities')
plt.show()
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 34/36
print(p_a_dishes)
print(p_a_radio)
print(p_a_phonecall)
F_onewayResult(statistic=0.6150287800038885, pvalue=0.5412256844691998)
F_onewayResult(statistic=2.211383605174631, pvalue=0.11112227149093898)
F_onewayResult(statistic=1.63970308504325, pvalue=0.19556825223405488)
F_onewayResult(statistic=0.4827571963186801, pvalue=0.6175019597524363)
Based on the p-values obtained, there appears to be no statistical evidence that time spent on the activities has difference between the living
environments.
Since there's multiple simultaneous tests, we should apply multiple correction again.
# Use Bonferroni correction
p_values = [p_a_cooking[1], p_a_dishes[1], p_a_radio[1], p_a_phonecall[1]]
ssm.multipletests(p_values, method='bonferroni')[1]
array([1. , 0.44448909, 0.78227301, 1. ])
After correcting the p-values, we can see that there's no signi cant difference when comparing to the p-values obtained from the test.
5. Exploring which of the activities are associated with each other
Next we use another way of estimating and visualising correlations between activities using matrixes. First by using pearson correlation
coe cient and then by using spearman's rank correlation coe cient.
# Identify pearson correlation between activities
activity_corr = data[numerical_columns].corr()
print("Activity Correlation Matrix:\n", activity_corr)
# Visualize the correlation matrix
sns.heatmap(activity_corr, annot=True, cmap='coolwarm')
plt.title('Correlation Between Activities')
plt.show()
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 34/36
Activity Correlation Matrix:
Cooking Washing dishes Listening to radio Phonecall
Cooking 1.000000 0.495309 0.098799 0.072268
Washing dishes 0.495309 1.000000 0.012370 0.172762
Listening to radio 0.098799 0.012370 1.000000 -0.019309
Phonecall 0.072268 0.172762 -0.019309 1.000000
The relationship between activities with Pearson correlation:
Cooking and Washing dishes: The correlation value is 0.5, indicating a moderate positive relationship. This means that as cooking time
increases, dishwashing time also tends to increase.
Listening to radio: Listening to the radio has a very low correlation with most other activities. This may re ect that radio listening time
does not depend much on other activities
Phone call: The correlation between phone calls and cooking is 0.1, which is very low, suggesting that there is no relationship between
phone calls and cooking time. The negative correlation between phone calls and museum visits (-0.01), suggests that people who spend
more time on calls tend to visit museums less, but the magnitude is very weak.
# Identify spearman correlation between activities
activity_corr = data[numerical_columns].corr(method='spearman'
print("Activity Correlation Matrix:\n", activity_corr)
# Visualize the correlation matrix
sns.heatmap(activity_corr, annot=True, cmap='coolwarm')
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 35/36
Cooking Washing dishes Listening to radio Phonecall
Cooking 1.000000 0.495309 0.098799 0.072268
Washing dishes 0.495309 1.000000 0.012370 0.172762
Listening to radio 0.098799 0.012370 1.000000 -0.019309
Phonecall 0.072268 0.172762 -0.019309 1.000000
The relationship between activities with Pearson correlation:
Cooking and Washing dishes: The correlation value is 0.5, indicating a moderate positive relationship. This means that as cooking time
increases, dishwashing time also tends to increase.
Listening to radio: Listening to the radio has a very low correlation with most other activities. This may re ect that radio listening time
does not depend much on other activities
Phone call: The correlation between phone calls and cooking is 0.1, which is very low, suggesting that there is no relationship between
phone calls and cooking time. The negative correlation between phone calls and museum visits (-0.01), suggests that people who spend
more time on calls tend to visit museums less, but the magnitude is very weak.
# Identify spearman correlation between activities
activity_corr = data[numerical_columns].corr(method='spearman'
print("Activity Correlation Matrix:\n", activity_corr)
# Visualize the correlation matrix
sns.heatmap(activity_corr, annot=True, cmap='coolwarm')
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 35/36
Activity Correlation Matrix:
Cooking Washing dishes Listening to radio Phonecall
Cooking 1.000000 0.536558 0.114889 0.052429
Washing dishes 0.536558 1.000000 0.054925 0.168064
Listening to radio 0.114889 0.054925 1.000000 0.050967
Phonecall 0.052429 0.168064 0.050967 1.000000
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 36/36
Cooking Washing dishes Listening to radio Phonecall
Cooking 1.000000 0.536558 0.114889 0.052429
Washing dishes 0.536558 1.000000 0.054925 0.168064
Listening to radio 0.114889 0.054925 1.000000 0.050967
Phonecall 0.052429 0.168064 0.050967 1.000000
11/1/24, 12:01 AM Project002.ipynb - Colab
https://colab.research.google.com/drive/1yypWPF1JaeBPrBM8L4KWQltdHLVHi_vQ#printMode=true 36/36